Thin Steel Plate Surface Rust Recognition Using Processing Light Measurement for Reduction of Laser Cutting Defect False Recognition

Recently, laser cutting is an essential technology for high-speed flexible sheet-metal processing. The problem of defective cutting has occurred even with appropriate cutting parameter settings. The authors have constructed a defect recognizer that uses data of light generated during thin plate cutting, and have achieved a high recognition rate. On the other hand, some misrecognition occurred, and all of the misrecognized data were found to be workpieces with rust on the surface. Therefore, as a method to reduce misrecognition, rust information should be acquired and used for defect recognition before cutting. This study aims to acquire rust information on the workpiece surface by sensing and to recognize the existence of rust in order to reduce the false recognition in thin plate cutting defect recognition. The proposed method consists of three steps. In the first step, the surface of the workpiece is irradiated by a low-power laser, and the light generated is measured using a spectrometer installed in the laser head. In the second step, the acquired spectral data is converted into a spectrogram, and the image is binarized using Otsu’s binarization method to obtain features. In the final step, a one-class support vector machine recognizes the existence of rust on a workpiece surface from the extracted features. Verification tests using a normal and two kinds of rusted surface plates data confirmed that the proposed method accurately detected the existence of rust. (Precision = 0.89, Recall = 1.0.) It was also confirmed that the low-power laser irradiation trace did not affect the spectral data of the cutting for defect recognition.